“Under the European climate law, EU countries must cut greenhouse gas emissions by at least 55% by 2030. Their goal is to make the EU climate neutral by 2050.”

• Need to identify effective climate policy (as economists, we love efficiency)
• Limited resources: time and money
• Dissonance between targets and policies
• Need to evaluate policy in a non-biased way

What does all of this mean for Austria?

Standard policy evaluation

• Identify effects-of-causes of single, known policies
• Difficult to isolate individual policies in a real-world setting
• Narrow analysis: potential to miss policies

Reverse-causal policy evaluation

• An agnostic approach to policy evaluation for policy mixes
• Identify a-priori unknown or underappreciated interventions

An application of Koch et al. (2022)

• “Attributing agnostically-detected large reductions in road \(CO_2\) emissions to policy mixes”
• What do I do differently? Focus on Austria, on all sectors
• Identify structural breaks in emissions, not accounted for by GDP or population, using machine learning
• Attribute breaks to policies, using emissions policy databases

Structural break identification

• \(CO_2\) emissions: combination of EDGAR (Emissions Database for Global Atmospheric Research) and International Energy Agencies (IEA) databases
• Population and GDP: World Bank, World Development Indicators

Policy databases

• The IEA’s Policies and Measures Database: past, existing, or planned climate and energy policies. Data is collected from governments, international organisations, and IEA analyses, and governments can review the provided information periodically.
• IEA/IRENA Renewable Energy Policies and Measures Database: a joint database of renewable energy policies and measures of the IEA and IRENA.
• The National Communications to the UNFCCC secretariat: obligatory for EU countries to submit regularly.

Structural break identification

• Two-way fixed effects (TWFE) panel estimators
• 26 time periods and 15 countries = 390 indicators (more than observations)
• 2 samples: EU15 and EU27 + Norway, Iceland, Switzerland, and the United Kingdom because they were part of the European Single Market and subject to harmonized regulations)
• Sparse treatment of countries using block search machine learning algorithm
• General-to-specific (GETS) model

General model

Specific model

Table 1. Negative structural breaks in Austrian emissions

Emissions per capita by sector

Table 2. Policy attribution to identified structural breaks

Headline

There appear to be very few highly effective climate policies identified using the reverse-causal approach for Austria from 1995 to 2021.

Caveats

• The reverse-causal approach is not a substitute, but a complement to standard policy evaluation
• This approach identifies relatively large effects
• This will not identify EU-wide regulations
• Causal interpretation of policies relies on assumption of no other interventions being present at the time of the break and on the exogeneity of the shock

EU15:

Austria, Belgium, Germany, Denmark, Spain, Finland, France, United Kingdom, Ireland, Italy, Luxembourg, Netherlands, Greece, Portugal, Sweden

EU31:

Austria, Belgium, Germany, Denmark, Spain, Finland, France, United Kingdom, Ireland, Italy, Luxembourg, Netherlands, Greece, Portugal, Sweden, Croatia, Bulgaria, Cyprus, Czechia, Estonia, Hungary, Lithuania, Latvia, Malta, Poland, Romania, Slovak Republic, Slovenia, Switzerland, Iceland, Norway

Table 3. Negative structural breaks in Austrian emissions (Level 1)

Table 4. Negative structural breaks in Austrian emissions (Level 2)

Table 5. Negative structural breaks in Austrian emissions